New technologies for monitoring the Common Agricultural Policy
The ECA has undertaken an audit on the use of new imaging technologies to monitor the Common Agricultural Policy (CAP). In this article Richard Hardy* Principal Manager in the ECA’s directorate responsible for auditing the CAP, provides some details about our audit and the latest developments in the use of technologies to monitor and control CAP spending. He also discusses the use of imaging technologies and digital audit techniques in the ECA’s wider audit work on the CAP.
By Richard Hardy, Sustainable Use of Natural Resources Directorate
Checks by monitoring: a major change in the management and control of the CAP
The Common Agricultural Policy (CAP) has a long history of using aerial photographs and satellite images to check some of its area-based support. Currently, paying agencies in the Member States use a computerised geographic information system of agricultural land parcels (the ‘Land Parcel Identification System’ — LPIS), which is based on aerial photographs and satellite images. The paying agencies use the LPIS to verify that they pay aid only for eligible agricultural land and only once for a given area of agricultural land.
The paying agencies update their LPIS imagery in general every three years. The LPIS does not show the crops or activities taking place on a given land parcel during the year (planting, harvesting, mowing, etc.). Therefore, the paying agencies have had to carry out field inspections for a sample of around 5 % of farmers, which is both time-consuming and costly. Since 1992, paying agencies have been able to adopt an alternative approach for inspecting agricultural parcels via high-resolution satellite images, called ‘checks with remote sensing’. This approach is less costly, but still requires human intervention in the form of operators to interpret satellite images, using computer-assisted photo-interpretation.
Since June 2015, the Sentinel 1 and 2 satellites, launched under the EU’s Copernicus programme, have been providing freely available high-resolution images. As these images are taken frequently, automated processing of time series data throughout the growing season makes it possible, without human intervention, to identify crops and monitor certain agricultural practices on individual parcels (see graph at Figure 1). Different crops reflect light in different ways, which is measured by the ‘vegetation index’ (the y axis in the graph). In this way, the variety of crop (e.g. wheat, maize) and agricultural activities taking place during the season, such as harvesting and mowing, can be identified.
Figure 1 — Example of a time profile of Sentinel-derived information showing
Changes in the vegetation cover for different crops
Since May 2018, Member States can use such automated processes based on Copernicus Sentinel data for continuous monitoring of the whole population of aid recipients, rather than only checking samples of farmers. This new approach is called ‘checks by monitoring.’ In 2018, Italy used it in the province of Foggia (Puglia). In 2019, 15 paying agencies in Belgium, Denmark, Italy, Malta and Spain used the new approach for some of their area-based payment schemes.
The processing of large quantities of data (e.g. through machine learning) provides the paying agencies with information on crop types and agricultural activity. They then assess the parcels using a system of coloured flags (see Picture 1) and can carry out targeted field visits as necessary, improving the efficiency of their controls.
Picture 1 — Simulation of possible outcome of parcels’ assessment
The European Commission, paying agencies, other stakeholders and experts agree that this approach represents a major change in the management and control of the CAP, with the following key benefits:
- increasing compliance, by helping farmers to meet aid scheme requirements through ongoing communication on potential non-compliance with specific requirements;
- reducing administrative burden and improving cost-effectiveness both for farmers and paying agencies;
- improving farm management, increasing farmers’ economic benefits and helping to reduce the negative environmental impacts of some agricultural activities.
ECA audit of the use of new imaging technologies to monitor the CAP
The ECA is currently finalising an audit to assess the EU’s use of new imaging technologies to monitor the CAP. As we indicated in the audit preview published in August 2019, we examined whether the Commission and Member States have taken sufficient steps to unlock the potential benefits of new imaging technologies for CAP monitoring. We assessed whether the Commission effectively encouraged widespread use of the new technologies and whether Member States had taken adequate action to deploy them. We also sought to identify examples of good practice in the use of new technologies for CAP monitoring, as well as shed light on obstacles hindering their wider deployment.
An assessment of the progress made is especially relevant now, as the results of our audit could be applied in the post‑2020 CAP. The increased use of new technologies for CAP monitoring may affect the future audit approach taken by national and EU audit bodies.
The new imaging technologies included in the scope of our audit are Copernicus Sentinel satellite data, images taken by drones, and geotagged images. Our audit did not cover the use of Copernicus Sentinel data for smart farming applications, crop yield forecasts or areas outside the CAP. We performed the audit fieldwork from April to September 2019 and expect to publish the special report at the end of January 2020.
During the audit, we organized a panel with experts from research institutes, universities, paying agencies, private companies and the European Environment Agency. The panel was very helpful in better understanding the impact of the new technologies, the progress made in their use, as well as the challenges for their wider deployment. We also organized a seminar for ECA staff on the use of new technologies for CAP monitoring. The seminar included presentations on relevant EU research projects, paying agencies’ experiences, and the potential of new technologies for monitoring climate and environment commitments, which led to lively discussions about the advantages and drawbacks of the use of new technologies in the CAP.
Perspectives for auditing the CAP with new technologies
In line with the ECA strategy, we are looking to develop digital audit techniques in our financial and compliance audit work, contributing to the ECA’s yearly Statement of Assurance (SoA), work on the CAP. We increasingly use the orthophotos and satellite imagery available in the paying agencies, together with other digital solutions (e.g. Google Earth), in order to assess the eligibility of land declared by farmers. This has enabled us to avoid on-site visits to farms in many cases, saving resources and increasing efficiency. In the 2019 SoA, we expect to obtain sufficient audit evidence from orthophotos and satellite imagery for more than half of our sample of direct payments to farmers under the European Agricultural Guarantee Fund.
We have also initiated a pilot project to examine the possible application of process mining techniques in our SoA work. This project is being carried out in cooperation with a Member State paying agency (see Box 1 below).
Box 1 — Pilot project: Process mining
In collaboration with colleagues from ECALab, the ECA’s centre for research and innovation, we are undertaking a pilot project in process mining in order to analyse the processing of aid applications and payments to beneficiaries at a selected Member State regional paying agency. Building on ECALab’s previous experiences with process mining, we will examine to what extent currently available commercial and open-source process mining software can be used for audit purposes. As process mining is one of the candidate technologies for the audit of entire populations of transactions, we will focus on the processing steps of all aid applications and payments in the paying agency within a specific timeframe, with special attention to checks, controls and segregation of duties within the process. The project will require close cooperation between the ECA and the paying agency, especially during the data acquisition phase, and to ensure the correct interpretation of the processing exceptions we may find during the analysis.
There are also possibilities to use digital techniques for performance audit. Some of the new and emerging technologies, such as natural language processing and semantic document search could allow us to rapidly process non-structured textual data relating to aid applications, tenders and procurements. The use of exogenous ‘big data,’ such as the automated collection and analysis of online news articles, scientific papers and public documents can also provide supplementary or corroborating audit evidence for performance audit projects.
All in all, we believe that the ever-increasing volume of new data available in a digital format, will be a major step forward for our audit of the CAP, both for assessing compliance and performance aspects. This is not only a question of making savings in the use of audit resources, but will also allow us to engage in full population audits, rather than concentrating on samples of transactions. We will be able to focus on outliers and other ‘red flag’ cases. These developments should not only contribute to further improving the financial management of the CAP, but also farm management as such, including the climate and environmental aspects related to agricultural activities. The new data and techniques, in combination with our traditional ‘analogue’ fieldwork, makes auditing EU’s agricultural expenditure more varied and interesting than ever.
*He acknowledges Els Brems, Jindrich Dolezal and Paulo Braz from the new imaging technologies audit team, Zsolt Varga from ECALab and Ramona Bortnowschi (private office attaché) for their most valuable input to this article.
This article was first published on the 1/2020 issue of the ECA Journal. The contents of the interviews and the articles are the sole responsibility of the interviewees and authors and do not necessarily reflect the opinion of the European Court of Auditors.
